(A) L- and M-cones are placed on a hexagonal grid (with noise to jitter the exact locations) with an appropriate density for the eccentricity of the model (5 deg here). The L:M cone ratio is 1:1, and the assignment of code ID (L or M) is random.

(B) The function of Horizontal cells is modeled as a resistive mesh where each conductance takes a value, g_h. The signals from the cone photoreceptors are input to the mesh through conductances, g_p. Each node has a capacitance, C, which plays a rold in determining the dynamics of the network. P_ij and H_ij are the values of the photoreceptor and horizontal cell signals, respectively.

(C) Bipolar (BP) and ganglion cell (RGC) signals are computed for each cone photoreceptor by simply subtracting the (weighted) value of H from P at each location (i,j). The BP signal is then converted to a spike rate to generate an inhomogeneous Poisson spike train, which represents the output of the RGC.